tion learning in locust (Forman, 1984; Hoi-ridge, 1962; Tosney and Hoyle, 1977), odor .... Testing always began after at least 2 hr of light and individuals were.
The Journal of Neuroscience May 1986, 6(5): 13251331
Classical Conditioning I. lzja Lederhendler,
Serge
of Hermissenda:
Gart, and Daniel
Origin of a New Response
L. Alkon
Section on Neural Systems, Laboratory of Biophysics, IRP, National Institute of Neurological and Communicative Disorders and Stroke, National Institutes of Health at the Marine Biological Laboratory, Woods Hole, Massachusetts 02543
Training of the marine snail Hermissenda crassicornis with paired light and rotation was previously shown to result in acquisition and retention of a behavioral change with many features characteristic of vertebrate associative learning. Here, this behavioral change is demonstrated to be classical, Pavlovianlike conditioning. A new response to light is formed (the CR) that is pairing-specific and resembles the unconditioned response (UCR) to rotation. The conditioned and unconditioned responses are relatively rapid, occurring within seconds of the onset of light or rotation stimuli, and correspond to pairingspecific reductions in speed during the same time period. Since the CR is independent of the presentation of rotation, and it is also expressed by the same effector system (the foot) responsible for the UCR, light stimulation has assumed some of the functional character of rotation.
modifications could not be produced by repeated presentations of the conditioned stimulus (CS) or of the unconditioned stimulus (UCS) (Crow and Alkon, 1978). There is no persistent suppression of phototaxis that might be interpreted as sensitization, habituation, or pseudoconditioning (Farley and Alkon, 1985). We know that at least some of the behavioral changes contributing to the overall reduction in positive phototaxis occur shortly after the test light is presented. One manifestation of this is the increased latency to initiate locomotion toward the light source (Crow and Offenbach, 1983; Farley and Alkon, 1982). Even though the time to initiate locomotion (measured as a latency to reach a criterion distance) is on the order of several minutes, it is clear from other observations (Barnes and Lederhendler, 198 1; Lederhendler and Alkon, 1984; Lederhendler et al., 1983; I. Lederhendler and D. L. Alkon, unpublished observations) that some changes happen immediately on presentation of the light. Individuals do move and locomote in very dim light during the time prior to reaching the criterion distance by which the start of locomotion has been measured (Crow and Offenbach, 1983; Farley and Alkon, 1982). This means that even in the least illuminated regions of the gradient, individuals experience changing levels of illumination. Thus, the increased latencies to “initiate” locomotion may be the consequence of a fundamental change in the responses of associatively trained Hermissenda to contrasting levels of light intensity (Lederhendler and Alkon, 1984). Furthermore, in at least one identified visual pathway (Goh and Alkon, 1982, 1984) training-dependent reductions in motor output within the first 5 set of light stimulation were directly correlated with the increased latency to enter the light (Goh et al., 1985; Lederhendler et al., 1982). In associative learning with Hermissenda, light has been used as the CS. In the present study, we have defined one response that occurs within 6 set of light onset: foot-lengthening. Furthermore, in this paper we show that in tubes, in the absence of light, the response to rotation (the UCR) is a reduction in the length of the foot. Alkon (1974) originally noted that Hermissenda placed in open beakers containing seawater responded to turbulence by increasing contact with the substrate through a “clinging” response. One feature of this clinging appeared to be a configurational change of the foot (the single organ of locomotion) as it increased its contact with the substrate. We present evidence that after training with paired light and rotation stimuli, a new response to light occurs that resembles the UCR-a decrease in length of the foot. These findings suggest that both of the essential characteristics of classical conditioning as a model of associative learning may be satisfied here (Gormezano, 1984; Gormezano and Moore, 1969). The CR (light-elicited foot-shortening) is independent of the presentation of the UCS. And, in addition, the CR is elicited by the same effector system (the foot) that elicits the UCR. The light therefore appears to have taken on some functional character of the rotation.
The development of invertebrate models of associative learning and memory depends on a compromise between nervous systems accessible to state-of-the-art analyses and learning behavior with features identified for vertebrate species. Some of the promising invertebrate models include instrumental leg-position learning in locust (Forman, 1984; Hoi-ridge, 1962; Tosney and Hoyle, 1977), odor aversion learning in Limax (Sahley et al., 198 la, b), chemosensory aversion learning in Pleurobranchaea (Mpitsos and Davis, 1973; Mpitsos et al., 1978) and gilland siphon-withdrawal in Aplysia (Carew et al., 198 1, 1983; Lukowiak and Sahley, 198 1). For Hermissenda, associative learning has been measured as reduction in positive phototaxis (Alkon, 1974). Pairing-specific increased latencies to enter the illuminated portion of a lightintensity gradient have been measured on days following experience with paired light and rotation stimuli (Crow and Alkon, 1978; Farley and Alkon, 1982; Harrigan and Alkon, 1985). Whereas during baseline tests before training, naive animals enter the light within several minutes, after training with paired light and rotation, these same animals may not enter for up to 2 hr. Individuals trained with random occurrences of the same number of light and rotation stimuli do not change significantly relative to their pretraining baselines. Aside from this pairing specificity, Hermissenda associative learning satisfies a host of criteria traditionally applied to vertebrate conditioning: stimulus specificity (Crow and Offenbach, 1983; Farley and Alkon, 1982) long-term retention (Crow and Alkon, 1978; Harrigan and Alkon, 1985) extinction (Richards et al., 1984) savings (Crow and Alkon, 1978), and contingency sensitivity (Farley, in press). Particularly important, nonassociative behavioral Received June 17, 1985; revised Aug. 30, 1985; accepted Sept. 10, 1985. We wish to acknowledge Bob and Linda Golder for preparation of the figures and Jeanne Kuzirian for typing the manuscript. We are grateful to Dr. I. Gormezano for stimulating discussions leading to this study, and to Dr. A. Kuzitian for his comments. Correspondence should be addressed to I. I. Lederhendler. 0270-6474/86/051325-07$02.00/O
1325
1326
Lederhendler et al.
Materials and Methods General methods Adult Hermissenda crassicornis 2-4 cm in length were provided by M. Morris (Sea Life Supply, Sand City, CA) and D. Anderson (Charleston, OR) and maintained individually at 13°C in a recirculating aquarium. The aquarium was filled with 50 gal of fresh filtered (100 PM) sea water. Animals were maintained in this system for 4-5 d before an experiment began. They were fed 20-40 mg of tunicate viscera daily. The animals were maintained on a diurnal light cycle of 12 hr light : 12 hr dark. Testing always began after at least 2 hr of light and individuals were always returned to the aquarium before the light went off. The intensity of light from a modified 40 W fluorescent tube that reached the animals in the aquarium was approximately 5 x 10-l WW cm-2. All light measurements were made with or calibrated to a radiometer (Yellow Springs Instrument Co., Yellow Springs, OH; Model 65A).
Vol. 6, No. 5, May 1986
Training After baseline responses were obtained, usually over a 2 d period, individuals were randomly assigned to each of the treatment groups: Paired, Random, or Naive. For the Paired group, training consisted of 50 trials a day for 3 d in which 30 set of diffuse white light (1.3 x 1OZw.W cm-2 in the periphery) was presented at random intervals (0.5-4 min), together with 30 set of rotation (97 rpm). Maximum rotation was estimated to begin about 500 msec after the onset of light. The Random group was presented with the same total number of light and rotation stimuli but on separate random programs such that no consistent association between them occurred. In fact, the program that determined random stimulus presentations contained 25-33% of trials in which the light and rotation stimuli coincided by chance alone. The presentation of light and rotation was controlled by an automated timing circuit (Tyndale and Crow, 1979). The Naive group was retested after 3 or 4 d of standard maintenance. Paired and Random animals were tested 16-28 hr after the conclusion of training.
Photography The principle that guided the design of the photographic techniques described here was to obtain clear images of the foot under conditions of varying illumination (e.g., before and after stimulation with light) without needing to adjust camera settings. This was accomplished by using film sensitive to red light and filtering the strobes and subilluminating lights with red filters. We used a motor-driven Nikon FM2 35 mm camera with a 55 mm Micro-Nikkor lens, fitted with a Schott RG-630 and Tiffen polarizing and dichroic filters. Four Vivitar 283 flash units with VP-l Varipower modules and two Tensor high-intensity subilluminatine liehts. all fitted with Schott RG-665 and Tiffen nolarizing and dich&“filters, were used to illuminate the subject. Some of the experiments used three 6.6 W fluorescent tubes filtered with a Kodak No. 2 safelight and polarizing filters for subillumination. The camera and flash units were mounted on a modified motorized x-y plotter to manually track the moving animal. The shutter release was triggered manually or by a photoelectric cell linked to a delay circuit. We used Kodak Recording 2475 film throughout. Time was monitored with a digital chronometer mounted in the camera field of view. This system provided clear images of Hermissenda in the “dark,” during rotation, as well as in the light, without the need to adjust camera settings. Intracellular recordings from Type B photoreceptors, darkadapted for 10 min and stimulated with the filtered flash units, showed no generator potential or increased discharge frequency for at least 60 set after the flash.
Responseto rotation (UCS) Animals were placed in a 1.4 cm diameter Plexiglas tube filled with filtered seawater. The tube was mounted on a motorized turntable 25 cm in diameter 15 cm below the camera. The animals were restrained to the end of the tube and dark-adapted to the red subilluminating light for 7 min. At this time, the restraint was removed and the position of the tube was adjusted so that the foot of the animal was toward the camera. The animals began to locomote in the dark, and several baseline photographs were obtained at the rate of one per second. Rotation, at 97 rpm, was started and photographs were obtained during the next 10 sec.
Responseto light (CS) The techniques for obtaining pictures of the response to light were the same as those described in the previous section. A fiberoptic bundle, mounted 10 cm above the center of the tube, and fitted with an optical diffuser was fed from a single 150 W tungsten halogen source. This directed illumination along the length of the tube to form a horizontal gradient 2.3 x lo3 r.~Lwcmm2 at the center and 0.2 x lo2 pW cmm2 at the periphery. Animals were dark-adapted to the dim red subillumination for at least 6 min while restrained to the end of the tube. The restraint was removed and the position of the tube adjusted so that the foot of the animal was toward the camera. After the animal had begun locomotion, several photographs were taken to provide baseline measurements in the dark. When the subject reached a preset position 20 cm from the center of the turntable, the light was turned on. The intensity of illumination reaching the animal at this point was 1.3 x lo2 PW cm-*. One picture was always obtained simultaneously with the onset of the light to provide a calibration point for stimulus duration. Two to four pictures were obtained during the next 6 set of light stimulation.
Measurement and statistics After the end of training and testing, the negatives were mounted as slides. The slides were coded to hide identification of the treatment groups until after measurements were obtained. Slides were projected onto a Photooptical Digitizer (L-W International, Woodland Hills, CA, model 110) for direct scaled measurements of the foot. Length measurements for each slide were scaled to a standard 2 cm calibration mark etched onto the test tube. Each measurement was repeated seven times and the mean (+-SD) of these measurements recorded. If the SD was greater than kO.005 cm the measurement was repeated and the set of values with the smaller SD was used. The greatest length of the foot in the dark before light stimulation was used as the standard referent value. During rotation, the lengths after 1, 3, 6, and 9 set of stimulation were subtracted from the prerotation baseline referent to provide a difference score. Isochronal points could be obtained from individual to individual because the turntable speed acted as a precise timer. Pictures of the response to light were obtained manually so that the time intervals between pictures varied * 1 sec. To estimate the lengths in response to light, therefore, we used the mean of the smallest and largest values obtained during the first 6 set of light stimulation. We also chose to monitor displacement in response to light adjusted for varying time intervals. This measure (speed) was obtained during comparable intervals before and after training for each individual within 6 set before and after light onset. We hoped the measure would help assess the relationship between foot shortening and other measures of phototaxis (particularly latency measures) that have been used previously. The findings of previous studies (Crow and Alkon, 1978; Crow and Offenbach, 1983; Farley and Alkon, 1982) showed that, after training, Paired animals locomote more slowly than either the Random or Naive animals. We wanted to see if this finding held during the initial seconds of light stimulation. It should be noted that this is a direct measurement of changes in speed in response to light, not a latency to initiate locomotion, because the animals were already locomoting in the dark. Several criteria were applied to the photographic images to determine their acceptability for inclusion in the sample. Two of these were technical, and one monitored the well-being of test animals. Individuals were judged as “not thriving” if their dark baseline lengths after training were significantly shorter (p < 0.05; t test) than before training. Nine individuals were excluded on this basis. This selection occurred before identification of their treatment group. The two technical selection criteria dealt with the positions and configurations of the animals, which we could not control once the photography began. To assure valid measurements, the entire image of the foot had to be within the central two-thirds of the long axis of the tube. In addition, the entire length of the foot had to be in contact with the substrate. This was determined directly from the photographs as portions of the perimeter that were not in focus. These technical difficulties resulted in numerous individuals whose responses we could not measure. Between-group differences in changes in length were assessed with a one-way analysis of variance (Winer, 1971). A posteriori t tests were used to compare means directly. Non-parametric comparisons were based on Siegel (1956).
Classical Conditioning of Hermissenda
The Journal of Neuroscience
0 TIME (set)
2
4 FROM
6 START
8 OF
IO
f
ROTATION
z
-0.l5L BEFORE TRAINING
Figure 1. Unconditioned response of the foot of Hermissenda elicited
by rotation(97rpm)in the dark.Time intervalsmonitored1,3,6, and 9 setafterthe startof rotationwerecomparedto thosejust prior to the startof rotation. Results
Figure 1 showsthe changein foot length as a function of the duration of rotation in the dark. The responseitself is illustrated photographically in the bottom panel of Figure 3. The length ofthe foot decreasedin all 20 animalstested.Mean length(*SD) in the dark just prior to rotation was 2.91 + 0.49 cm. Average increasesin foot length were 14.8% at 1 set into rotation (n = 20), 13.1%after 3 set (n = 20), 11.1%after 6 set (n = 4), 3.7% after 9 set (n = 4). The samplesize diminished rapidly after 6 set asa resultof a combination of configurational and positional changeswith continued rotation. Usually, this meant the axis of the foot shifted too far to one sideof the tube. But the results indicate that foot shortening is greatest within 6 set after the start of rotation and subsequentlyappearsto diminish. The three groups of animals whose responsesto light were tested did not differ in baselinelengths of the foot (maximum length in the dark). Mean lengths (cm f SD) were as follows: Before training-Paired, 2.78 f 0.32; Random, 3.14 f 0.69; Naive, 2.85 + 0.32. After training-Paired, 3.05 + 0.35; Random, 3.27 + 0.67; Naive, 3.06 f 0.44. Table 1 showsthat before training 19 out of the sampleof 25 individuals respondedto the onset of light by lengtheningthe foot. The binomial probability associatedwith such a distribution occurring by chance is p = 0.007. This small but quite reliablelengtheningis illustrated photographically in the upperleft panel of Figure 3. No between-groupdifferenceswere found before training in the changesin foot length in responseto light (Fig. 2). After training with either paired or random presentationsof light and rotation stimuli, significantbetween-groupdifferenceswerefound in the changesof foot length in responseto light (F(2,22) = 5.6 1, p < 0.05). A posterioricomparisonsof the meansindicated that Paired animals were significantly different from both the Random(t(14) = 2.98,~ < O.Ol)andNaive(t(l5) = 2.95,~ < 0.01) groups. The latter two groups were not significantly different from each other. Examination of Table 1 showsthat all the paired animals showeda reduction in foot length in response to light stimulation after training. And, in fact, only the Paired group showeda significantchangeto a shorteningresponseafter training compared to before training (McNemar Test of the significanceof changes;x2 = 7.00, p c 0.01). Changesin speedin the dark and within 6 set of presentation
1327
CONDITIONED RESPONSE
UNCONDITIONED RESPONSE
Figure 2. Classicalconditioningof a new response to light in Hermissenda. A, Beforetraining,foot lengthincreased in response to light
onset.Random@)andNaive(Cl)groupscontinueto showlight-elicited lengthening, but Paired@) animalsbecame conditionedto shortenthe lengthof the foot in response to a light stimulus. B, Unconditioned response after6 setof rotation.TheCR isabout28%of theUCR.Note
the difference in scale between A and B.
of the light were assessedas a consequenceof training. No significant between-group differenceswere found for speedin the dark before training (Paired, 2.03 f 0.30 mm/set; Random, 1.90 f 0.44 mm/set; Naive, 1.89 +- 0.30 mm/set). Similarly, speedsin the light before training did not differ significantly (F(2,20) = 2.57) betweenthe groups(Paired, 2.30 + 0.75 mm/ set; Random, 1.67 + 0.56 mm/set; Naive, 2.29 + 0.47 mm/ set). However, sincethe Random group did appearto be slower, post hoc paired comparisonswere applied, and they supported the original conclusionof no differences.In any case,an intrinsically lower Random group would have a conservative influenceon the directional hypothesisthat the Paired animalsbecome slowerthan Random or Naive animals. In fact, asFigure 4 illustrates, the mean relative changein speedin the light, but not in darkness,wassignificantly different betweenthe treatment groups(F(2,20) = 4.2 1, p < 0.05). A posteriori paired comparisons of the meansshowedthat the Paired animals were significantly slower than either the Random (t(12) = 2.18, p < 0.05) or Naive (t(14) = 2.45, p < 0.05) animals. Theseresults indicate that after training, Paired animalsexhibit a new responseto light- shorteningratherthan lengthening
Table1. Frequency of occurrence of increased or decreased lengths of the foot of Hermissenda to light onset
Treatment
Before Training Number Number increased decreased
After Training Number Number increased decreased
Paired Random Naive Total
7 5 7 19
0 5 6 11
1 3 2 6
8 3n 3b 14
~These individuals were different from those that decreased in length before training. * One of these individuals decreased in length of the foot before training.
vol. 6, No. 5, May 1986
Lederhendler et al.
1328
BEFORE CONDITIONING
CONDITIONED RESPONSE
TllYE DARK
IN THE LIGHT
\
shortening
UNCONDITIONED RESPONSE
Figure 3. Photographic representation of classically conditioned responses of Hermissenda foot. Bottom panel, An overlay of two photographs taken in the dark; solid white line, the outline of the foot 1 set before rotation; dashed line, outline of the foot after 3 set of rotation at 97 rpm. Upper panels, Comparisons of lengths in the light to lengths in the dark before and after conditioning with paired light and rotation stimuli. C,, Length in the dark before training. &, Length in the light before training ( C, > C, = lengthening). Cg,Length in the dark during retention of learned behavior. P,, Length in the light during retention ( !, > & = shortening). & Length during rotation was always smaller than before rotation began. the foot. This responsehappenswithin the first 6 set of light
stimulation. During this same6 set period, thesesameanimals, as a group, exhibit a significantly reduced speedin the light. Discussion In Hermissendathe experienceof pairings of light and rotation stimuli has the effect of forming a new responseto light (the CR)-foot-shortening-that resemblesthe responseto rotation (the UCR). However, sincelocomotion in Hermissenda,as in
gastropodsgenerally, dependson muscular, secretory, and ciliary processes, foot-shorteningneednot occurexclusively through muscular contractions. Increasedadhesionof the mucus or reduced ciliary beating might produce similar effects. The description of classicallyconditioned featuresof associative learning in Hermissendabecamepossiblewith the specification of an objective measureof the UCR. There may be other UCRs to rotation, or to different intensities of rotation stimuli, that we have not yet examined. These other measures
The Journal
Classical
of Neuroscience IN THE DARK
IN THE
PRN
P R N
Conditioning
of Hermissenda Liaht Liqht
LIGHT
Gravity,
Receptor Organs c
5
s E -40
5s
Optic Ganglion C
-30 i
-50 L
Cerebropleural Ganglion
Figure 4. Percentage changein speed in thelight.andin thedarkduring
retention,ascomparedto beforeconditioning.Speeds werecalculated from photographicmeasurements of displacement 6 set beforeand 6 setafter light onset.The reductionin speedin the Pairedanimals(P) wassignificantly greaterthanin theRandom(R) or Naive(N)treatments in response to light stimulation,but not to dark. may or may not show classicalconditioning. However, the accumulated evidence to date (Alkon, 1984; Alkon et al., 1982, 1985; Crow and Alkon, 1980; Farley et al., 1983; Goh et al., 1985; Lederhendlerand Alkon, 1984; West et al., 1982) indicates that primary and intrinsic conductance changesin the medial Type B photoreceptor, which persistafter training, initiate a cascadeof effects within different identifiable neuronal pathwaysthat modify the expressionof at leastseveraldifferent visually dependent behaviors. Specific conditioning-induced output changesin the photoreceptorsneed not be expressedin all visual behaviors. Rather, different componentsof the photoreceptor output can influence various behaviors, depending on which neural pathways are excited and the circumstancesof the behavioral observations. It is important to emphasizethat the ionic conductancechanges have been related to Hermissendavisual behavior after dark adaptation and for an identified Type B photoreceptor-the medial B cell. Crow’s (1985b) report of reduced depolarization and impulseactivity is related to different stimulus conditions (after 5 min of light adaptation) and for unidentified Type B photoreceptors. It is significant, therefore, that in the present study we have demonstratedbehavioral changeswithin seconds of the onsetof light stimulation. Conditioned foot-shortening may mark the beginning of a seriesof learnedmodifications of visual responses.This series includesa delay in the time to start locomotion at the dim end of the gradient (Crow and Offenbach, 1983; Farley and Alkon, 1982) reducedspeedof locomotion while moving toward the center of the gradient (Farley and Alkon, 1982), reduced responseto the contrast differences that make up the gradient (Lederhendlerand Alkon, 1984;and unpublishedobservations), increasedlatencies to enter the brightest part of the gradient (Crow and Alkon, 1978;Crow and Offenbach, 1983;Farley and Alkon, 1982;Goh et al., 1985; Harrigan and Alkon, 1985),and, finally, reduced preference for the more intense areasof illumination (Crow, 1985a;Lederhendler et al., 1980). Modifications that may be primary have been identified in two of these behaviors: they occur early and they may greatly influence the other responsemeasures.Thesemodifications are conditioned foot-shortening, as reported in the present study, and the reducedability to detect contrast differencesin the gradient (Lederhendlerand Alkon, 1984;and unpublishedobservations).Both are especiallynoteworthy becausethey match closely the time courseof cellular correlates.
Turning -Orientation
Foot Contraction -Clinging
Figure 5. The flow of visualand graviceptiveinformationand their learnedinteractionsmay be distributedthroughidentified(solid lines) andpredicted(dashed lines)neuralpathwaysin sucha waythat different and even oppositebehavioralresponses will result.Convergence of informationaboutthe simultaneous occurrence of light androtationis knownto occurat the receptorlevel,the optic ganglion,andin interneuronsfoundin thecerebropleural ganglion.Pairing-specific increases in the activity of Type B photoreceptors may leadto foot-contraction in an excitatory pathway (open triangles) throughcpg intemeurons (Akaike and Alkon, 1980).In a test situationrequiringresponses to contrast,reducedorientationcapabilitiesmay be producedvia an inhibitory pathway(solid triangles) throughthe medialType A photoreceptors(Goh et al., 1985;Lederhendler andAlkon, 1984).
For example, one neural pathway has been identified (Goh and Alkon, 1984) that can account for the disruption of Hermissenda’sability to respondto contrast differencesin the gradient (Lederhendler and Alkon, 1984), and the consequentincrease in latencies to enter the light (Goh et al., 1985; Lederhendler et al., 1980). In this pathway, pairing-specific changesin the output of the network (putative motomeuron, MNl) were measuredduring the first 5 set of light stimulation (Goh et al., 1985). Impulse activity in MN1 was found to be inversely related to firing frequency in the medial Type B photoreceptor. The inverse relationship occurs becausethe medial Type B cell inhibits the medial A photoreceptor, which, in turn, excites identifiable intemeurons presynaptic to MN 1. After training with paired light and rotation, enhancedexcitability in the Type B photoreceptorwascorrelatedwith reducedexcitation in MN 1 in responseto light. Reducedexcitation in MN 1causes lessturning of the animal toward the stimulated side (Goh and Alkon, 1984). But the intemeuronal relationshipsdescribedfor the turning pathway do not also account for the neural control of the behavioral changesdescribedin the presentstudy. However, consideration of the behavioral observationstogether may provide certain guidelines for studiesof cellular correlates of classical conditioning of Hermissenda.In the presentstudy, foot-lengthening in responseto light stimulation is associatedwith forward movement, but not with the initiation of locomotion, since
1330
Lederhendler
locomotion was already in progresswhen the light was presented. Turning toward the light, as discussedabove, is also associatedwith forward movement. Thus, we can distinguish an orientation systemthat involves approachbehavior and tuming movements. On the other hand, foot-shortening is an opposite responseassociatedwith clinging, stopping, and withdrawal behaviors which seemto disrupt locomotion. The organization of behaviors mediated by visual and graviceptive input canbe summarizedinto three functional systems: (1) An arousal system in which light and rotation initiate locomotion; we know this system is sensitive to light intensity, sensoryadaptation, and diurnal factors (Barnesand Lederhendler, 1981; Crow, 1985a; Crow and Offenbach, 1983; Lederhendleret al., 1980).(2) An orientation systemin which turning and forward (approach)locomotion would dependon a balance of inputs to the left or right sideand, at the sametime, minimal input to the foot-contraction pathway. (3) A systemrelating to foot-shortening, reduced forward locomotion, withdrawal, or stopping.This would be a high-threshold systemresponsiveto intenserotation or turbulence. Thus, for example, in the presenceof strong bilateral rotational stimuli, the foot-contraction pathway would be activated and override turning, forward movement, and lengthening. Visual information may therefore be distributed through the nervous systemto allow for expressionof different behavioral responses.Figure 5 outlines identified neural connections that provide a working model to account for the two major response systemsof interest in the presentstudy: the orientation system and the foot-shortening system. The turning pathway is activated via the Type A photoreceptor and modulated by inhibitory input from the Type B cell. It is also excited directly by hair-cell input (Goh and Alkon, 1984). The foot-contraction pathway, on the other hand, is activated by excitatory input from the Type B photoreceptor. Interneurons of this type, which are also excited directly by the hair cells, were described by Akaike and Alkon (1980). References
Akaike, T., and D. L. Alkon (1980) Sensory convergence on central visual neurons in Hermissenda. J. Neurophvsiol. 44: 501-5 13. Alkon, D. L. (1974) Associative training in Hermissenda. J. Gen. Physiol. 64: 70-84. Alkon, D. L. (1984) Calcium-mediated reduction of ionic currents: A biophysical memory trace. Science 226: 1037-1045. Alkon, D. L., I. Lederhendler, and J. Shoukimas (1982) Primary changes of membrane currents durina retention of associative learning. Science 215: 693-695. Alkon, D. L., M. Sakakibara, R. Forman, J. Harrigan, I. Lederhendler, and J. Farlev (1985) Reduction of two voltage-dependent K+ currents mediates retention of a learned association. Behav. Neural Biol. 44: 278-300.
Barnes, E. S., and I. I. Lederhendler (1981) Dark-adaptation effects on photobehavior of Hermissenda crassicornis (Gastropoda: Nudibranchia). Biol. Bull. 159: 479 (Abstr.). Carew, T. J., E. T. Walters, and E. R. Kandel (1981) Classical conditioning in a simple withdrawal reflex in Aplysia californica. J. Neurosci. I: 1426-1437. Carew, T. J., R. D. Hawkins, and E. R. Kandel (1983) Differential classical conditioning of a defensive withdrawal reflex in Aplysia callfirnica. Science 219: 397-400. Crow, T. (1985a) Conditioned modification of phototactic behavior in Hermissenda. I. Analysis of light intensity. J. Neurosci. 5: 209214. Crow, T. (1985b) Conditioned modification of phototactic behavior in Hermissenda. II. Differential adaptation of B-photoreceptors. J. Neurosci. 5: 2 15-223. Crow, T. J., and D. L. Alkon (1978) Retention of an associative behavioral change in Hermissenda. Science 201: 1239-l 24 1. Crow, T. J., and D. L. Alkon (1980) Associative behavioral modification in Hermissenda: Cellular correlates. Science 209: 412-414. Crow, T., and N. Offenbach (1983) Modification of the initiation of
et al.
Vol. 6, No. 5, May 1986
locomotion in Hermissenda: Behavioral analysis. Brain Res. 271: 301-310. Farley, J. (in press) Contingency learning and causal detection in Hermissenda: Behavioral and cellular mechanisms. Behav. Neurosci. Farley, J., and D. L. Alkon (1982) Associative and behavioral change in Hermissenda: Consequences of nervous system orientation for lightand pairing-specificity. J. Neurophysiol. 48: 785-807. Farley, J., and D. L. Alkon (1985) Cellular mechanisms of learning, memory, and information storage. Annu. Rev. Psychol. 36: 4 19-494. Farley, J., W. G. Richards, L. Ling, E. Liman, and D. L. Alkon (1983) Membrane changes in a single photoreceptor during acquisition cause associative learning in Hermissenda. Science 221: 1201-1203. Forman, R. (1984) Leg position learning by an insect. I. A heat avoidance learning paradigm. J. Neurobiol. IS: 127-140. Gart, S., I. Lederhendler, and D. L. Alkon (1983) An infrared macrophotographic technique for quantifying the behavioral response to rotation ofthe gastropod Hermissenda crassicornis. Biol. Bull. (Abstr.) 165: 525.
Goh, Y., and D. L. Alkon (1982) Convergence of visual and statocyst inputs on interneurons and motorneurons of Hermissenda: A network design for associative conditioning. Sot. Neurosci. Abstr. 8: 825. Goh, Y., and D. L. Alkon (1984) Sensory, intemeuronal and motor interactions within Hermissenda visual pathway. J. Neurophysiol. 52: 156-169. Goh, Y., I. Lederhendler, and D. L. Alkon (1985) Input and output changes of an identified neural pathway are correlated with associative learning. J. Neurosci. 5: 536-543. Gormezano, I. (1984) The study of associative learning with CS-CR paradigms. In Primary Neural Substrates of Learning and Behavioral Change, D. L. Alkon and J. Farley, eds., pp. 5-24, Cambridge U.P., Cambridge, U.K. Gormezano, I., and J. W. Moore (1969) Classical conditioning. In Learning: Processes, M. H. Marx, ed., pp. 121-203, Macmillan, New York. Harrigan, J. F., and D. L. Alkon (1985) Individual variation in associative learning of the nudibranch mollusc Hermissenda crassicornis. Biol. Bull. 168: 222-238. Horridge, G. A. (1962) Learning leg position by the ventral nerve cord of headless insects. Proc. R. Sot. London IBiol.1 157: 33-52. Lederhendler, I., and D. L. Alkon (1984) Reduced withdrawal from shadows: An expression of primary neural changes of associative learning in Hermissenda. Sot. Neurosci. Abstr. 10: 270. Lederhendler, I. I., E. S. Barnes, and D. L. Alkon (1980) Complex responses to light of the nudibranch Hermissenda &as&ornis (Gastronoda: Onisthobranchia). Behav. Neural Biol. 28: 2 18-230. Lederhendler: I., Y. Goh, and D. L. Alkon (1982) Type B photoreceptor changes predict modification of motoneuron responses to light during retention of Hermissenda associative conditioning. Sot. Neurosci. Abstr. 8: 825. Lederhendler, I., S. Gart, and D. L. Alkon (1983) Associative learning in Hermissenda crassicornis (Gastropoda): Evidence that light (the CS) takes on characteristics of rotation (the UCS). Biol. Bull. (Abstr.) 165: 529.
Lukowiak, K., and C. Sahley (198 1) The in vitro classical conditioning of the gill withdrawal reflex ofAplysia californica. Science 212: 15 161518. Mpitsos, G. J., and W. J. Davis (1973) Learning: Classical and avoidance conditioning in the mollusc Pleurobranchaea. Science 180: 3 17320.
Mpitsos, G. J., S. D. Collins, and A. McClellan (1978) Learning: Model system for physiological studies. Science 199: 497-506. Richards, W., J. Farley, and D. L. Alkon (1984) Extinction of associative learning in Hermissenda: Behavior and neural correlates. Behav. Brain Res. 14: 161-170. Sahley, C., A. Gelperin, and J. W. Rudy (198 la) One-trial associative learning modified food odor preferences of a terrestrial mollusc. Proc. Natl. Acad. Sci. USA 78: 640-642. Sahley, C., J. W. Rudy, and A. Gelperin (198 lb) An analysis of associative learning in a terrestrial mollusc. I. Higher-order conditioning, blocking, and a transient US pre-exposure effect. J. Comp. Physiol. 144: l-8. Siegel, S. (1956) Non-parametric Statistics for the Behavioral Sciences, McGraw-Hill, New York. Tosney, T., and G. Hoyle (1977) Computer-controlled learning in a simple system. Proc. R. Sot. London [Biol.] 195: 365-393.
The Journal of Neuroscience
Classical Conditioning of Hermissenda
Tyndale, C. L., and T. Crow (1979) An IC control unit for generating random and nonrandom events. IEEE Trans. Biomed. Engin. BME 26: 649-655. West, A., E. Barnes, and D. L. Alkon ( 1982) Primary changes of voltage
1331
responses during retention of associative learning. J. Neurophysiol. 48: 1243-1255. Winer, B. J. (197 1) Statistical Principles in Experimental Design, McGraw-Hill, New York.
